Abstract

Immune cells enter the central nervous system (CNS) from the circulation under normal conditions for immunosurveillance and in inflammatory neurologic diseases. This review describes the distinct anatomic features of the CNS vasculature that permit it to maintain parenchymal homeostasis and which necessitate specific mechanisms for neuroinflammation to occur. We review the historical evolution of the concept of the blood-brain barrier and discuss distinctions between diffusion/transport of solutes and migration of cells from the blood to CNS parenchyma. The former is regulated at the level of capillaries, whereas the latter takes place in postcapillary venules. We summarize evidence that entry of immune cells into the CNS parenchyma in inflammatory conditions involves 2 differently regulated steps: transmigration of the vascular wall into the perivascular space and progression across the glia limitans into the parenchyma.

Introduction

The interface between the immune and nervous systems has long fascinated both immunologists and neuroscientists. This is best exemplified at the most prominent interface between these systems, the blood-brain barrier (BBB). In discussions of clinical and experimental inflammatory brain diseases, brain homeostasis, and psychology, the BBB takes on almost mystical significance. It is striking, however, how many different versions of the BBB one can encounter in such discussions, depending on the focus of the discussants. We and others have reviewed this topic in other forums but felt it would be useful to focus on the anatomic and physiologic intricacies of the BBB with reference to its pathobiologic functions. We emphasize the concept that cellular transmigration across the BBB is not a single-step phenomenon as occurs in other tissues, but that it involves 2 distinct processes that overlap within the perivascular space. This 2-step nature of cellular migration into the central nervous system (CNS) imposes specific constraints that are fundamental to understanding many facets of neuroimmunology and CNS pathophysiology.

Morphology of the Neurovascular Unit: Vascular Wall, Perivascular Space, and Glia Limitans

The topography and architecture of CNS microvessels are designed for 2 major site-specific tasks. First, because neuronal activity within the CNS strictly depends on homeostasis, the microvasculature of the brain must prevent ions and other solutes in the blood from entering the parenchyma. Second, to enhance regional blood flow wherever it is needed, the vascular tree has to have the capacity to respond rapidly to neuronal activity. The first task is reflected by the term BBB; the second is referred to as “neurovascular coupling.” Both the differentiation of BBB-typical endothelial cells and the response of the vessels to enhanced electric activity seem to depend in various ways on astrocyte signaling provided at a unique structure, the glia limitans (Fig. 1). The glia limitans is composed of astrocyte processes and covers the entire surface of the brain and spinal cord on the external face, toward the subarachnoid space (glia limitans superficialis), and internally around the vessels (glia limitans perivascularis; Fig. 2A).

FIGURE 1.

Glia limitans and perivascular spaces. (A) Immunostaining for glial fibrillary acidic protein highlights the “glia limitans perivascularis” in a human brain. (B) In standard histologic sections, perivascular spaces are not apparent around capillaries (arrows) but are evident (asterisk) where capillaries merge to form postcapillary venules.

FIGURE 1.

Glia limitans and perivascular spaces. (A) Immunostaining for glial fibrillary acidic protein highlights the “glia limitans perivascularis” in a human brain. (B) In standard histologic sections, perivascular spaces are not apparent around capillaries (arrows) but are evident (asterisk) where capillaries merge to form postcapillary venules.

FIGURE 2.

(A) Schematic diagram of different anatomic regions from brain parenchyma to skull. Superficial vessels of the brain [6] are located in the subarachnoid space [7]. This compartment is delineated by the arachnoid mater [4] and the pia mater [3]. The surface of the brain is completely covered by the astrocytic endfeet of the glia limitans [2]. Toward the subarachnoid space, these endfeet are designated as glia limitans superficialis (A); toward the vessels inside of the brain, they are termed glia limitans perivascularis (B). On their way from the surface to the deep areas of the brain, the vessels take leptomeningeal connective tissue with them, thereby forming perivascular spaces [10], which remain connected to the subarachnoid space. 1 indicates perikaryon of an astrocyte; 2, glia limitans superficialis; 3, connective tissue of the pia mater (inner layer of the leptomeninges); 4, arachnoid mater (outer layer of the leptomeninges); 5, subarachnoid connective tissue (trabeculae arachnoideae); 6, subarachnoid vessel; 7, subarachnoid space; 8, dura mater (pachymeninges); 9, neurothelium; 10, perivascular space; 11, penetrating vessel; 12, capillary; 13, glia limitans perivascularis. (B, C) Ultrastructure of the PVS, which is bordered by the other walls of a postcapillary venule and the glia limitans (arrowheads) or their basement membranes. The field corresponds to the white bar (A). (C) Higher magnification of the field depicted (B) shows the basement membranes. In noncapillary vessels, at least 3 basement membranes can be distinguished: the endothelial membrane (dotted line), the outer vascular membrane (dashed line), and the membrane of the glia limitans (dotted and dashed). (D) Ultrastructure of the region corresponding to the black bar (A). (D) In capillaries (12 in [A]), the basement membranes are merged to form a “fused gliovascular membrane” that occludes the perivascular space. (E) Higher magnification of the field depicted (D). The capillary wall consists of endothelium [E], endothelial basement membrane (dotted line), and Pe. The fused gliovascular membrane is shown by a continuous black line. It is directly apposed to the glia limitans. Astrocyte processes of the glia limitans are indicated by arrowheads. The arrows (C, E) point to tight junctions between endothelial cells. The overlap of adjacent endothelial cells is a hallmark of the BBB and is evident in the capillary, but not in the venule. E, endothelial cell; L, lumen; MC, mural cell, an intermediate form between smooth muscle cell and pericyte; Pe, pericytes; PVS, perivascular space. Figure reprinted from Radivoj V. Krstic: Die Gewebe des Menschen und der Säugetiere (Human and Mammalian Tissues), 1988, with kind permission of Springer Science + Business Media.

FIGURE 2.

(A) Schematic diagram of different anatomic regions from brain parenchyma to skull. Superficial vessels of the brain [6] are located in the subarachnoid space [7]. This compartment is delineated by the arachnoid mater [4] and the pia mater [3]. The surface of the brain is completely covered by the astrocytic endfeet of the glia limitans [2]. Toward the subarachnoid space, these endfeet are designated as glia limitans superficialis (A); toward the vessels inside of the brain, they are termed glia limitans perivascularis (B). On their way from the surface to the deep areas of the brain, the vessels take leptomeningeal connective tissue with them, thereby forming perivascular spaces [10], which remain connected to the subarachnoid space. 1 indicates perikaryon of an astrocyte; 2, glia limitans superficialis; 3, connective tissue of the pia mater (inner layer of the leptomeninges); 4, arachnoid mater (outer layer of the leptomeninges); 5, subarachnoid connective tissue (trabeculae arachnoideae); 6, subarachnoid vessel; 7, subarachnoid space; 8, dura mater (pachymeninges); 9, neurothelium; 10, perivascular space; 11, penetrating vessel; 12, capillary; 13, glia limitans perivascularis. (B, C) Ultrastructure of the PVS, which is bordered by the other walls of a postcapillary venule and the glia limitans (arrowheads) or their basement membranes. The field corresponds to the white bar (A). (C) Higher magnification of the field depicted (B) shows the basement membranes. In noncapillary vessels, at least 3 basement membranes can be distinguished: the endothelial membrane (dotted line), the outer vascular membrane (dashed line), and the membrane of the glia limitans (dotted and dashed). (D) Ultrastructure of the region corresponding to the black bar (A). (D) In capillaries (12 in [A]), the basement membranes are merged to form a “fused gliovascular membrane” that occludes the perivascular space. (E) Higher magnification of the field depicted (D). The capillary wall consists of endothelium [E], endothelial basement membrane (dotted line), and Pe. The fused gliovascular membrane is shown by a continuous black line. It is directly apposed to the glia limitans. Astrocyte processes of the glia limitans are indicated by arrowheads. The arrows (C, E) point to tight junctions between endothelial cells. The overlap of adjacent endothelial cells is a hallmark of the BBB and is evident in the capillary, but not in the venule. E, endothelial cell; L, lumen; MC, mural cell, an intermediate form between smooth muscle cell and pericyte; Pe, pericytes; PVS, perivascular space. Figure reprinted from Radivoj V. Krstic: Die Gewebe des Menschen und der Säugetiere (Human and Mammalian Tissues), 1988, with kind permission of Springer Science + Business Media.

By definition, capillaries in all organs lack a tunica media of smooth muscle cells. Capillary endothelial cells in the CNS are surrounded by large numbers of pericytes that are embedded into the vascular basement membrane, which is directly attached to the glia limitans (Fig. 2C). This contrasts with CNS arteries and veins in which the distance between the endothelial layer and the glia limitans is much larger because there are various layers of intervening smooth muscle cells. In postcapillary CNS venules, there is an intermediate situation with a perivascular space and the presence of “mural cells” within the endothelial basement membrane (Fig. 2B); these have characteristics of pericytes or vascular smooth muscle cells (1).

As they pass from the subarachnoid space into the deep areas of the brain, the large vessels are surrounded by a connective tissues consisting of leptomeningeal mesothelial cells, macrophages, and other antigen-presenting cells comprising the cellular components of the perivascular space (Fig. 2A). Thus, perivascular spaces are located between the basement membrane of the vascular wall and the basement membrane of the glia limitans and are continuous with the subarachnoid space (Fig. 1B). In the capillary segment, the basement membranes contain laminins 411 and 511, whereas in the glia limitans, they contain laminins 111 and 211; the latter fuse to form a “gliovascular membrane” that closes the perivascular space (Fig. 2D) (2). As a consequence, a unidirectional flow of extracellular fluids from the parenchyma along perivascular spaces into the subarachnoid space can be driven by the arterial pulse; this substitutes for the lack of lymphatic drainage in the brain (3).

Where/What is the BBB?

The term BBB describes the phenomenon that was first observed by Paul Ehrlich (4) in the course of his studies with “intravital dyes.” These hydrophilic compounds can be injected into the blood and change their color depending on their oxidative state. Ehrlich's goal was to compare the oxygen consumption of different organs based on the color they exhibited upon dye injection. A prerequisite of this approach was that the dye should be equally distributed within the body at the beginning of the experiment, but Ehrlich noted that this was not the case. In particular, the brain exhibited little or no staining. He therefore concluded that the brain lacks affinity for these dyes. Continuing with this work, Biedl and Kraus (5) injected bile acids into the CNS, and the Berlin physician Lewandowski (6) injected sodium ferrocyanide into the CNS, from which he concluded that the “capillary wall must block the entrance of certain molecules” not normally present in the blood. This was the first time that a special barrier function was postulated for brain capillary endothelial cells.

The development of electron microscopy allowed further focus on the “barrier.” Reese and Karnovsky (7) were the first to describe the unique belt-like tight junctions in brain capillary endothelial cells, which they regarded as correlates of a BBB. In addition, they noted “a paucity” of transportation vesicles that linked the concept of a barrier to more dynamic processes. Indeed, we now know that in addition to the mechanical barrier maintained by tight junctions at the capillary level, a number of permanently active transportation mechanisms contribute to excluding blood molecules that could harm the parenchymal milieu necessary for neural transmission (8).

The term BBB thus provides a metaphorical explanation for a phenomenon first noted by Ehrlich rather than pointing to a single anatomic structure or mechanism. In fact, much confusion comes from attempts to translate the concept of a BBB into the field of cellular neuroinflammation. Whereas the permission or restriction of passage of molecules from the blood is a major function of capillaries, immune cells are recruited to the CNS at the level of postcapillary venules; the architecture and topographic relationships of these vessels to the brain parenchyma are strikingly different due to the existence of mural cells and perivascular spaces. The width of perivascular spaces is variable in the healthy CNS (Fig. 1B). Distance from the glia limitans seems to affect astrocyte signaling on endothelial cell differentiation; and therefore, notable differences have been described in the organization of tight junctions in capillaries compared with precapillary and postcapillary vessels (Figs. 2C, E) (9). Unfortunately, a systematic ultrastructural study of the molecular composition of tight junctions in capillaries versus in postcapillary venules has not yet been performed. It is not generally appreciated that it is only at first glance that markers of the BBB such as horseradish peroxidase (HRP) do not leave the blood stream and enter the brain when they are injected into the circulation (10). A closer look reveals that HRP staining can be observed in the vascular wall of larger vessels and in macrophages of the perivascular spaces and the choroid plexus, which thus act as substitutes of BBB function at the precapillary and postcapillary levels; their capacity to act as phagocytic scavengers compensates for the enhanced endothelial layer permeability in these areas of the vascular tree. This additional level of regulation of diffusion from the blood into the parenchyma of the brain prompted Goldmann (11) to propose the notion of a “physiologic bordering membrane” consisting of highly phagocytic macrophages in 1913. Since then, the penetration of BBB markers through noncapillary brain vessels has remained the neuroanatomists' “dirty little secret” (12, 13). Nevertheless, it can be concluded that the mechanisms that maintain a barrier for solutes in postcapillary venules involve additional components.

From studies of bone marrow chimeras, it became evident that the population of perivascular macrophages is regularly exchanged by hematogenous precursors in the absence of pathologic processes in the brain parenchyma (14-16). In light of the more recent appreciation of the effects of irradiation on cellular recruitment into the brain (17), it is noteworthy that this exchange also occurs in nonirradiated rats (18). Thus, monocytes can apparently cross the vascular wall of brain vessels under normal conditions. They do not, however, normally progress through the glia limitans into the neuropil, but are retained in perivascular spaces. Only when there is intraparenchymal damage such as axonal degeneration or inflammation, as in multiple sclerosis (MS) or its animal model experimental autoimmune encephalomyelitis (EAE), does the glia limitans become permissive for monocytes to migrate into the neuropil; some of them may then transform into microglia (15, 19, 20). For example, Babcock et al (21) showed that the chemokine monocyte chemotactic protein 1 (CC-ligand 2) produced by astrocytes and microglia is a critical factor driving the recruitment of monocytes into areas of Wallerian degeneration.

Two distinct steps can also be distinguished for T-cell migration into the CNS parenchyma. One group also demonstrated that the elimination of perivascular macrophages did not interfere with perivascular infiltration after the induction of EAE in mice, but the progression of T cells across the glia limitans was completely blocked (22). Thus, macrophages and/or antigen recognition seem to be required for T cells to perform the second step of migration. The Angelov group showed that sensitizing rats with HRP and subsequently injecting HRP into their subarachnoid space led to diffusion along perivascular spaces and phagocytosis of HRP by perivascular macrophages. When the animals were immunologically boosted with HRP, T cells entered the perivascular spaces but did not cross the glia limitans; they were essentially put “on hold” in the perivascular compartment (23). Similarly, deletion of both matrix metalloprotease (MMP) 2 and MMP-9 confers resistance to EAE by trapping leukocytic infiltrates in the perivascular space, suggesting that MMPs produced by macrophages are required for leukocyte infiltration across the glia limitans (24).

Thus, the term BBB in the context of infiltration of leukocytes into the CNS parenchyma can be confusing. Physiologically, the term refers to the vascular segment of the capillaries that regulate diffusion of solutes, whereas in an inflammatory response, the term refers to the postcapillary venules, that is, the vessels from which leukocytes migrate into the CNS. These are distinct vascular segments (25). For leukocytes to reach the CNS parenchyma, they need to perform 2 differently regulated steps: first, to cross the vascular wall, and second, to traverse the glia limitans (Fig. 3). Understanding the distinct molecular mechanisms is critical because it seems that activated lymphocytes regularly penetrate the endothelial barrier for immunosurveillance of the CNS, but only upon penetration of the glia limitans and infiltration of the CNS parenchyma do leukocytes come into direct contact with the parenchyma, which leads to clinical disease.

FIGURE 3.

The 2 steps to neuroinflammation. (A) In the first step of neuroinflammation (1st), circulating immune cells cross the microvascular central nervous system (CNS) endothelium to reach the PVS. In the second step (2nd), immune cells progress across the glia limitans to enter the CNS Pa. (B, C) The distinction between perivascular and intraparenchymal infiltration in standard histologic sections in the spinal cord of a mouse with EAE. (B) In a paraffin section, the border built by the glia limitans has been traversed by inflammatory cells on the right side (2nd), whereas cells are restricted to the perivascular space on the left side (1st). (C) Phase-contrast microscopy highlights the distinction. The arrowheads point to 2 cells that appear to “digest” or otherwise “open” the basement membrane of the glia limitans (white line), which is not visible at the light microscopic level. Pa, parenchyma; PVS, perivascular space.

FIGURE 3.

The 2 steps to neuroinflammation. (A) In the first step of neuroinflammation (1st), circulating immune cells cross the microvascular central nervous system (CNS) endothelium to reach the PVS. In the second step (2nd), immune cells progress across the glia limitans to enter the CNS Pa. (B, C) The distinction between perivascular and intraparenchymal infiltration in standard histologic sections in the spinal cord of a mouse with EAE. (B) In a paraffin section, the border built by the glia limitans has been traversed by inflammatory cells on the right side (2nd), whereas cells are restricted to the perivascular space on the left side (1st). (C) Phase-contrast microscopy highlights the distinction. The arrowheads point to 2 cells that appear to “digest” or otherwise “open” the basement membrane of the glia limitans (white line), which is not visible at the light microscopic level. Pa, parenchyma; PVS, perivascular space.

Migration Across the Vascular Wall and (Sometimes) Progression Across the Glia Limitans

As previously described, there are 2 principal physical barriers to cellular entry to the CNS parenchyma, the vascular endothelium and the glia limitans, each with its own basement membrane. The perivascular space between them represents not so much a barrier as a regulatory checkpoint that depends on the precise anatomic localization and the immune status of the patient or animal.

Migration of leukocytes across the vascular wall is governed by adhesion molecules, cytokines, chemokines, and their receptors. These are integrated in a sequential process of leukocyte-endothelial cell interaction that was first formulated by Butcher (26) and Dustin and Springer (27). The molecular mechanisms involved in the multistep paradigm of leukocyte recruitment across the vascular wall have been extensively reviewed (26-29). Briefly, leukocytes in the so-called “fast lane” of in the vessel lumens engage with inflamed endothelium via tethering and rolling (i.e. slowing) and are triggered by chemokine gradients to higher affinity adhesion, locomotion, and frank adhesion before they cross the vascular wall (diapedesis; Fig. 4).

FIGURE 4.

Multistep paradigm for leukocyte entry to the central nervous system (CNS). In the first step of neuroinflammation circulating, immune cells engage in a multistep interaction with CNS microvascular endothelial cells that they cross by either transcellular or paracellular routes. After passing through the endothelial basement membrane (EBM), they reach the perivascular space (PVS). After additional activation triggers that are probably provided by perivascular macrophages or dendritic cells, MMP-2 and MMP-9 are produced and allow the second-step immune cell entry across the glia limitans or pavenchymal basement membrane (PBM) into the CNS parenchyma.

FIGURE 4.

Multistep paradigm for leukocyte entry to the central nervous system (CNS). In the first step of neuroinflammation circulating, immune cells engage in a multistep interaction with CNS microvascular endothelial cells that they cross by either transcellular or paracellular routes. After passing through the endothelial basement membrane (EBM), they reach the perivascular space (PVS). After additional activation triggers that are probably provided by perivascular macrophages or dendritic cells, MMP-2 and MMP-9 are produced and allow the second-step immune cell entry across the glia limitans or pavenchymal basement membrane (PBM) into the CNS parenchyma.

CNS Address and Welcome Mats

Tissue-specific “area codes” are combinations of adhesion and signaling ligands that are expressed selectively or at higher levels in the CNS vasculature and are thought to favor interactions with leukocytes that express the specific selection of appropriate counterligands. The area code concept would explain selective migration of leukocytes to different anatomic regions such as the CNS, although the precise molecular composition of CNS area codes has not yet been defined. At present, there are no CNS-specific endothelial cell adhesion ligands that are absolutely specific. Certain combinations, however, can favor leukocyte interactions with cerebral or spinal cord endothelial cells. For example, whereas most circulating leukocytes express the adhesion molecule L-selectin ligand (CD62-L), a subset of activated or memory-effector T cells have downregulated L-selectin. Therefore, they are less likely to reenter lymph nodes from the circulation, and their migration may favor the CNS. The selectins and the selectin ligand P-selectin glycoprotein ligand 1 are not required for leukocyte infiltration across CNS microvessels such as in EAE (29, 30), and in mice, a counterligand for α4-integrins, vascular cell adhesion molecule 1, is constitutively expressed on spinal cord endothelial cells and mediates the capture and firm adhesion of T-cell blasts via α4-integrin (31,32).

Selectins are a family of carbohydrate-recognizing cell surface proteins (calcium-dependent C-type lectins) that are widely expressed both by leukocytes and endothelial cells. Selectin-driven tethering promotes rolling and permits the slowed-down leukocytes to be triggered by local gradients of immobilized chemokines. The process is selective in that those blood cells with appropriate chemokine receptors will then proceed to subsequent integrin activation (inside-out signaling), locomotion to sites of entry, and tight adhesion (28). The fact that selectins are not equally required for tethering adhesion in the inflamed CNS vessels influences the design of therapies directed against entry of potentially pathogenic leukocytes (29,33).

Integrin interactions mediate firm leukocyte adhesion and are therefore likely targets for therapy. The efficacy of α4-integrin-directed immunotherapy in MS represents an important proof-of-principle in this regard because it opens up possibilities that other adhesion molecule-ligand interactions may be useful drug targets (31). The recent identification of a role for the adhesion ligand-activated leukocyte cell adhesion molecule or CD166 in CD6+ CD4+ T-cell interactions with cerebral endothelial cells is a potential candidate for this therapeutic approach (34).

Integration of Autoimmune and Infectious Immune Principles

The multistep paradigm is also applicable to immune responses to CNS infections. In the skin, local virus infections induce the trafficking of antigen-presenting Langerhans cells to lymph nodes where, as dendritic cells, they induce a memory-effector T-cell response that is then directed to the site of infection via local inflammation (28). A role for local inflammation in this process is noteworthy. In contrast, EAE in rodents can be induced by adoptive transfer to an unmanipulated recipient of activated CD4+ T cells that have specificity for myelin peptides. Alternatively, it can be induced by subcutaneous immunization with these myelin peptides in adjuvant in susceptible rodent strains. Although many of the actively induced EAE models involve pertussis toxin, it is not needed for disease induction either by immunization or in adoptive transfer in certain rodent strains. It may, however, increase the disease incidence (35); thus, in these models, a CNS-specific disease can be induced without the use of adjuvant or preexisting local inflammation. One interpretation of these observations is that whether leukocyte interaction with CNS endothelium results in extravasation or disease may be determined by the sum of the molecular interactions between them. This presumes that highly activated T cells that have sufficiently elevated levels of adhesion molecules and cytokine/chemokine production can elicit responses from an otherwise quiescent vascular endothelium. In support of this concept, T-cell diapedesis across noninflamed retinal microvessels fails until the interaction has induced endothelial intercellular cell adhesion molecule 1 expression, which probably triggers the extravasation (36). Moreover, T-cell-derived tumor necrosis factor can elicit a chemokine response from BBB-associated cells, principally dendritic cells/macrophages/pericytes/microglial cells (37). The activation status of the T cell is critical to this process because only freshly activated T cells can transfer EAE and productively interact with the noninflamed BBB (32). The extent to which activated T cells and local inflammation (e.g. resulting from infection or other local CNS responses) may be involved in CNS inflammation in MS is not known.

In Vitro Models

In vitro models of CNS endothelium have resulted in many advances in our understanding of the BBB. Specific in vitro models mimic many aspects of live animals, but a full neurovascular unit is too complex to reproduce in vitro at present. Cocultures of cerebral endothelial cells with apposed astrocytes that induce a tight junctional barrier in vitro have been designed to model the capillary BBB and investigate the passage of solutes across this vascular bed (37). Recently, these systems have been adapted to model postcapillary venules where astrocytes and endothelia are not in intimate contact, that is, more closely mimicking postcapillary venules (32).

Exchange and Turnover of Cells in the Perivascular Space

Cellular migration across the BBB is recognized as a critical step for CNS inflammation. Seminal studies by Hickey and Kimura (14) showed that for induction of EAE, encephalitogenic T cells and bone marrow-derived cells in CNS, presumed to be at the BBB, should share expression of major histocompatibility complex (MHC), or be MHC-compatible. Further studies by Hickey et al (38) and by Sedgwick et al (39) determined that a population of perivascular macrophages is replaced in radiation chimeras, distinct from parenchymal microglia, which are not. This observation points to the ability of blood-derived cells to exchange within the perivascular space, that is, at least some blood-derived cells cross the vascular endothelium under normal circumstances. These would be the relevant MHC-matched cells with which T cells must interact. This is consistent with the observation that dendritic cells, which are usually found in extraparenchymal locations (e.g. the perivascular and subarachnoid spaces and meninges), are critical for the induction of EAE (40).

Is There Antigen-Driven Interaction Between T Cells and Endothelia?

It is uncertain whether endothelial cells present antigen to CD4+ T cells for the induction of CNS inflammation. The relative absence of expression of the CD4 ligand MHC II on normal endothelium and the fact that EAE can be adoptively transferred to naive recipients suggest that antigen-driven interactions between CD4+ T cells and endothelial cells are not necessary for their extravasation. The logic behind this argument contributes to interpretation of data (41, 42) showing that naive CD4+ T cells can access the CNS. Endothelial cells can be induced to express MHC II and costimulatory molecules in vitro, and this may identify the capacity of endothelial cells to present antigen. Some observations in situ, however, do not support this (43, 44), suggesting more cautious interpretation. Expression of such molecules was increased in disease-associated endothelial inflammation in autopsy material (45), but this may represent a post hoc effect relative to the initial T-cell entry that is the focus here. The observation that adhesion of CD4+ T cells is increased when activated endothelial cells express costimulatory molecules CD40, B7.1/CD80, or B7.2/CD86 (46, 47) likely has functional significance but does not prove antigen recognition in vivo.

A constitutive low level of the CD8 ligand MHC I on cerebral endothelium suggests that migrating CD8+ T cells may engage in antigen recognition (48). In that study, naive CD8+ T cells were detected in the side of the brain in which cognate peptide had been injected, and this was blocked by antibody against MHC I, which could only access the luminal endothelial surface (48). The authors' interpretation of the absence of T cells in the contralateral brain, in which an irrelevant antigen was injected, that CD8+ T cells could only enter if antigen was presented, is valid only if a transient “look-and-leave” entry can be excluded. The issue turns on whether egress or reverse migration contributes to net effects (see following).

Emigration

Egress or reverse migration is difficult to measure (49). One recent study described cells that were assumed to have left the CNS (50), but consideration of cellular exit from the CNS can confound interpretation of in vivo analyses. For example, the failure to detect T cells within the CNS cannot be taken to show that they did not enter the CNS. When T cells are detected, it is presumed to be the net result of entry plus either proliferation or accumulation (51). The concept of immunosurveillance predicts that potentially protective T cells should be able to access all tissues, which can be extended to include that in the absence of functional antigen presentation, they may either leave again or die in situ (veni, vidi, vici or veni, mori 52).

Transcellular Versus Paracellular Routes of Migration

Leukocytes may go through endothelial cells (transcellular route) or go between them (paracellular route). Transcellular migration was originally described by Marchesi and Gowans (53) in electron microscope studies that showed “emperipolesis,” whereby lymphocytes were enveloped by cells at vascular walls. Which molecular mechanisms trigger transcellular versus paracellular diapedesis is presently unclear; the use of both routes in different vascular beds has been described for both lymphocytes and neutrophils (53, 54). In the CNS, the passage between endothelial cells would necessitate the disruption of the complex tight junctions; molecular interactions with the junctional molecules claudins, occludin, junctional adhesion molecules, CD99, and Zonula occludens molecules would be involved (54). Although there is accumulating evidence for transcellular diapedesis of leukocytes across the BBB during inflammation (54), the molecular mechanisms remain to be investigated. The functional and structural characteristics.

TABLE.

Summary of Functional and Structural Characteristics of the Capillary and Post-capillary Blood-Brain Barrier*

Migration Across the Glia Limitans and Its Regulation

The glia limitans seems to be the site at which MMPs play the most significant role in leukocyte migration. Toft-Hansen et al (57) showed that, whereas a broad-spectrum MMP inhibitor did not affect the subclinical perivascular accumulation of leukocytes in a transgenic mouse expressing the chemokine CC-ligand 2 in the CNS, MMP inhibition blocked leukocyte entry into CNS parenchyma induced by pertussis toxin. This indicates that chemokine-driven crossing of the endothelium and perivascular leukocyte accumulation was not dependent on MMPs, whereas they were critical for migration across the glia limitans. The specific MMPs involved were not identified in that study. As previously mentioned, however, in EAE, MMP-2 and MMP-9 proteolytically cleave dystroglycan, which anchors astrocyte endfeet to the glia limitans basement membrane via binding to the extracellular matrix molecules agrin and perlecan (24). Moreover, staining for the transmembrane β-dystroglycan isoform is lost in the pertussis toxin-treated myelin basic protein promoter-driven CC-ligand 2 transgenic mice, indicating that dystroglycan breakdown occurs in this model (Füchtbauer et al, unpublished observation).

The glia limitans is not as accessible to in vitro analysis as the cerebrovascular endothelium. Nevertheless, there currently is a consensus that chemokines and integrins exert their major effects at the vascular endothelium, whereas MMPs have more prominent roles in glia limitans breakdown. Astrocytes are a potential source of chemokines at the BBB and probably contribute significantly to perivascular chemokine levels under physiologic conditions (58, 59). Another contribution that astrocytes make is modifying overall BBB permeability. Several groups have shown that permeability of artificial BBB endothelial monolayers to macromolecules is significantly decreased by coculture with astrocytes or by treatment with astrocyte-derived soluble factors (60, 61). In a recent study that showed expression of angiotensin receptors at the BBB in MS brain and on endothelial cells in vitro, Wosik et al (62) reported that astrocytes can modify expression and phosphorylation of the endothelial cell tight junction-associated molecule occludin via release of angiotensin. Rodent astrocytes have also been reported to express AT1 angiotensin receptors and may respond to vascular tension-regulatory hormones (63-65).

The Blood-Cerebrospinal Fluid Barrier as Possible Entry Site of Immune Cells into the CNS

In contrast to the classical endothelial BBB, the epithelial blood-cerebrospinal fluid (CSF) barrier of the choroid plexus epithelium has not, until very recently, been considered to be an entry site for immune cells or pathogens into the CNS (66). The choroid plexus extends from the ventricular surface into the lumen of the ventricles; its major known function is the secretion of CSF. It is organized in a villous surface, including an extensive microvascular network enclosed by a single layer of cuboidal epithelium (67). The microvessels within the choroid plexus parenchyma differ from those of the brain parenchyma because they allow free movement of molecules via fenestrations and intercellular gaps (68). The functional barrier between the blood compartment and the CSF is located at the level of the choroid plexus epithelial cells that form tight junctions that inhibit paracellular diffusion of water-soluble molecules (69). These tight junctions are the morphologic correlate of the blood-CSF barrier and have a unique molecular composition of transmembrane proteins with occludin, claudin-1 and claudin-2, and claudin-11 (69).

Regional Differences in CNS Microvessels: Parenchyma, Leptomeninges, and Circumventricular Organs

In addition to the choroid plexus, there are structures in the CNS of mammals that lack an endothelial BBB. These areas are commonly referred to as the circumventricular organs (CVOs) and perform homeostatic and neurosecretory functions; the neurons within them monitor hormonal stimuli and other substances within the blood or secrete neuroendocrines into the blood (70, 71). Circumventricular organs are localized at strategic points close to the midline of the brain within the ependymal walls lining the third and fourth ventricles. Because they lack an endothelial BBB, they lie within the blood milieu and thus form a blood-CSF barrier. Similar to the choroid plexus, a complex network of tight junctions connecting specialized ependymal cells (tanycytes) seal off the CNS from the CVOs (71, 72). Although the CVOs have often been referred to as “windows of the brain” with respect to soluble mediators, entry of immune cells into the brain via the CVOs across the tanycytic barriers has not been investigated extensively to date (73).

In addition to the obvious differences in fenestrated capillaries within the CVOs and the barrier-forming microvessels within the CNS, regional differences in the characteristics of the BBB have also been described. For example, mouse endothelioma cells established from the cerebellum respond in a more pronounced fashion than cerebral endotheliomas to proinflammatory cytokines by increased permeability and reduced transendothelial electrical resistance (74).

Molecular and cellular differences between meningeal and parenchymal microvascular beds have also been described (75, 76); meningeal microvessels lack astrocytic ensheathment. Interestingly, the differences between meningeal and parenchymal CNS blood vessel endothelial cells extend to differences in their expression patterns of P- and E-selectin. In the healthy CNS, meningeal blood vessel endothelial cells can be distinguished from those in the CNS parenchyma by their constitutive expression of P-selectin, which is absent from endothelial cells of parenchymal blood vessels (30, 77, 78). Furthermore, injection of proinflammatory cytokines into mice induced expression of E-selectin in meningeal but not in parenchymal CNS blood vessel endothelial cells (79). Thus, leukocytes that extravasate across meningeal vessels arrive in the leptomeningeal space from which they might travel within the perivascular spaces; if confronted with an inflammatory stimulus, they could enter the CNS parenchyma. Meningeal blood vessels have been described as being leakier than parenchymal blood vessels and more reactive to proinflammatory cytokines, and this might contribute to leukocyte recruitment restricted to the meningeal compartment.

Conclusion

In summary, we have proposed that the term BBB is somewhat of a misnomer. The barrier concept is more applicable to solute entry, the neuroinflammatory relevance of which relates more to edema than to cellular migration. The process of leukocyte entry into the CNS parenchyma is controlled by different cellular components at the level of postcapillary venules. Thus, we highlight the concept that regulation of cellular entry involves migration across 2 distinct structures, the vascular wall and the glia limitans and their associated basement membranes. This migration involves 2 steps for the development of neuroinflammation.

Acknowledgments

The authors acknowledge discussions within the COST Action BM0603 Inflammation in Brain Disease Neurinfnet and networking support from COST.

References

1.
Gerhardt
H
Betsholtz
C
.
Endothelial-pericyte interactions in angiogenesis
.
Cell Tissue Res
 
2003
;
314
:
15
23
2.
Sixt
M
Engelhardt
B
Pausch
F
Hallmann
R
Wendler
O
Sorokin
LM
.
Endothelial cell laminin isoforms, laminins 8 and 10, play decisive roles in T cell recruitment across the blood-brain barrier in experimental autoimmune encephalomyelitis
.
J Cell Biol
 
2001
;
153
:
933
46
3.
Carare
RO
Bernardes-Silva
M
Newman
TA
et al
.
Solutes, but not cells, drain from the brain parenchyma along basement membranes of capillaries and arteries: Significance for cerebral amyloid angiopathy and neuroimmunology
.
Neuropathol Appl Neurobiol
 
2008
;
34
:
131
44
4.
Ehrlich
P
.
Das Sauerstoff-Bedürfnis des Organismus. Eine farbenanalytische Studie. [On the oxygen consumption of the body. A study using intravital dyes.]
 .
Berlin
:
Verlag von August Hirschwald
,
1885
5.
Biedl
A
Kraus
R
.
Über eine bisher unbekannte toxische Wrikung der Gallensäuren auf das Zentralnervensystem
.
Zentralblatt Innere Medizin
 
1898
;
19
:
1185
200
6.
Lewandowski
M
.
Zur Lehre von der Cerebrospinalflüssigkeit
.
[On the cerebrospinal fluid]. Zentralblatt Klein Forschung
 
1900
;
40
:
480
94
7.
Reese
TS
Karnovsky
MJ
.
Fine structural localization of a blood-brain barrier to exogenous peroxidase
.
J Cell Biol
 
1967
;
34
:
207
17
8.
Zlokovic
BV
.
The blood-brain barrier in health and chronic neurodegenerative disorders
.
Neuron
 
2008
;
57
:
178
201
9.
Ge
S
Song
L
Pachter
JS
.
Where is the blood-brain barrier…really?
J Neurosci Res
 
2005
;
79
:
421
27
10.
Kristensson
K
Olsson
Y
.
Accumulation of protein tracers in pericytes of the central nervous system following systemic injection in immature mice
.
Acta Neurol Scand
 
1973
;
49
:
189
94
11.
Goldmann
E
.
Vitalfärbungen am Zentralnervensystem
.
Beitrag zur Physio-Pathologie des Plexus Choroideus und der Hirnhäute. [Intravital labelling of the central nervous system. A study on the pathophysiology of the choroid plexus and the meninges.] Abhandlungen der königlich preubischen Akademie der Wissenschaften Physikalisch-Mathematische Classe
 
1913
;
1
:
1
64
12.
Galea
I
Bechmann
I
Perry
VH
.
What is immune privilege (not)?
Trends Immunol
 
2007
;
28
:
12
18
13.
Janeway
CA
jr
.
Approaching the asymptote evolution and revolution in immunology
.
Cold Spring Harb Symp Quant Biol
 
1989
;
54
:
1
13
14.
Hickey
WF
Kimura
H
.
Perivascular microglial cells of the CNS are bone marrow-derived and present antigen in vivo
.
Science
 
1988
;
239
:
290
92
15.
Priller
J
Flugel
A
Wehner
T
et al
.
Targeting gene-modified hematopoietic cells to the central nervous system: Use of green fluorescent protein uncovers microglial engraftment
.
Nat Med
 
2001
;
7
:
1356
61
16.
Bechmann
I
Priller
J
Kovac
A
et al
.
Immune surveillance of mouse brain perivascular spaces by blood-borne macrophages
.
Eur J Neurosci
 
2001
;
14
:
1651
58
17.
Mildner
A
Schmidt
H
Nitsche
M
et al
.
Microglia in the adult brain arise from Ly-6ChiCCR2+ monocytes only under defined host conditions
.
Nat Neurosci
 
2007
;
10
:
1544
53
18.
Bechmann
I
Kwidzinski
E
Kovac
AD
et al
.
Turnover of rat brain perivascular cells
.
Exp Neurol
 
2001
;
168
:
242
49
19.
Bechmann
I
Goldmann
J
Kovac
AD
et al
.
Circulating monocytic cells infiltrate layers of anterograde axonal degeneration where they transform into microglia
.
FASEB J
 
2005
;
19
:
647
49
20.
Ladeby
R
Wirenfeldt
M
Garcia-Ovejero
D
et al
.
Microglial cell population dynamics in the injured adult central nervous system
.
Brain Res Brain Res Rev
 
2005
;
48
:
196
206
21.
Babcock
AA
Kuziel
WA
Rivest
S
Owens
T
.
Chemokine expression by glial cells directs leukocytes to sites of axonal injury in the CNS
.
J Neurosci
 
2003
;
23
:
7922
30
22.
Tran
EH
Hoekstra
K
van
Rooijen N
Dijkstra
CD
Owens
T
.
Immune invasion of the central nervous system parenchyma and experimental allergic encephalomyelitis, but not leukocyte extravasation from blood, are prevented in macrophage-depleted mice
.
J Immunol
 
1998
;
161
:
3767
75
23.
Walther
M
Popratiloff
A
Lachnit
N
et al
.
Exogenous antigen containing perivascular phagocytes induce a non-encephalitogenic extravasation of primed lymphocytes
.
J Neuroimmunol
 
2001
;
117
:
30
42.
24.
Agrawal
S
Anderson
P
Durbeej
M
et al
.
Dystroglycan is selectively cleaved at the parenchymal basement membrane at sites of leukocyte extravasation in experimental autoimmune encephalomyelitis
.
J Exp Med
 
2006
;
203
:
1007
19
25.
Bechmann
I
Galea
I
Perry
VH
.
What is the blood-brain barrier (not)?
Trends Immunol
 
2007
;
28
:
5
11
26.
Butcher
EC
.
Leukocyte-endothelial cell recognition: Three (or more) steps to specificity and diversity
.
Cell
 
1991
;
67
:
1033
36
27.
Dustin
ML
Springer
TA
.
Role of lymphocyte adhesion receptors in transient interactions and cell locomotion
.
Annu Rev Immunol
 
1991
;
9
:
27
66
28.
von
Andrian UH
Mackay
CR
.
T-Cell function and migration
.
Two sides of the same coin. N Engl J Med
 
2000
;
343
:
1020
34
29.
Engelhardt
B
Ransohoff
RM
.
The ins and outs of T-lymphocyte trafficking to the CNS: Anatomical sites and molecular mechanisms
.
Trends Immunol
 
2005
;
26
:
485
95
30.
Doring
A
Wild
M
Vestweber
D
Deutsch
U
Engelhardt
B
.
E- and P-selectin are not required for the development of experimental autoimmune encephalomyelitis in C57BL/6 and SJL mice
.
J Immunol
 
2007
;
179
:
8470
79
31.
Man
S
Ubogu
EE
Ransohoff
RM
.
Inflammatory cell migration into the central nervous system: A few new twists on an old tale
.
Brain Pathol
 
2007
;
17
:
243
50
32.
Vajkoczy
P
Laschinger
M
Engelhardt
B
.
Alpha4-integrin-VCAM-1 binding mediates G protein-independent capture of encephalitogenic T cell blasts to CNS white matter microvessels
.
J Clin Invest
 
2001
;
108
:
557
65
33.
Engelhardt
B
Kempe
B
Merfeld-Clauss
S
et al
.
P-selectin glycoprotein ligand 1 is not required for the development of experimental autoimmune encephalomyelitis in SJL and C57BL/6 mice
.
J Immunol
 
2005
;
175
:
1267
75
34.
Cayrol
R
Wosik
K
Berard
JL
et al
.
Activated leukocyte cell adhesion molecule promotes leukocyte trafficking into the central nervous system
.
Nat Immunol
 
2008
;
9
:
137
45
35.
Munoz
JJ
Bernard
CC
Mackay
IR
.
Elicitation of experimental allergic encephalomyelitis (EAE) in mice with the aid of pertussigen
.
Cell Immunol
 
1984
;
83
:
92
100
36.
Xu
H
Manivannan
A
Liversidge
J
Sharp
PF
Forrester
JV
Crane
IJ
.
Requirements for passage of T lymphocytes across non-inflamed retinal microvessels
.
J Neuroimmunol
 
2003
;
142
:
47
57
37.
Murphy
CA
Hoek
RM
Wiekowski
MT
Lira
SA
Sedgwick
JD
.
Interactions between hemopoietically derived TNF and central nervous system-resident glial chemokines underlie initiation of autoimmune inflammation in the brain
.
J Immunol
 
2002
;
169
:
7054
62
38.
Hickey
WF
Vass
K
Lassmann
H
.
Bone marrow-derived elements in the central nervous system: An immunohistochemical and ultrastructural survey of rat chimeras
.
J Neuropathol Exp Neurol
 
1992
;
51
:
246
56
39.
Sedgwick
JD
Schwender
S
Imrich
H
Dorries
R
Butcher
GW
ter
Meulen V
.
Isolation and direct characterization of resident microglial cells from the normal and inflamed central nervous system
.
Proc Natl Acad Sci U S A
 
1991
;
88
:
7438
42
40.
Greter
M
Heppner
FL
Lemos
MP
et al
.
Dendritic cells permit immune invasion of the CNS in an animal model of multiple sclerosis
.
Nat Med
 
2005
;
11
:
328
34
41.
Krakowski
ML
Owens
T
.
Naive T lymphocytes traffic to inflamed central nervous system, but require antigen recognition for activation
.
Eur J Immunol
 
2000
;
30
:
1002
9
42.
Bauer
J
Bradl
M
Hickey
WF
et al
.
T-Cell apoptosis in inflammatory brain lesions: Destruction of T cells does not depend on antigen recognition
.
Am J Pathol
 
1998
;
153
:
715
24
43.
Pedersen
EB
McNulty
JA
Castro
AJ
Fox
LM
Zimmer
J
Finsen
B
.
Enriched immune-environment of blood-brain barrier deficient areas of normal adult rats
.
J Neuroimmunol
 
1997
;
76
:
117
31
44.
Redwine
JM
Buchmeier
MJ
Evans
CF
.
In vivo expression of major histocompatibility complex molecules on oligodendrocytes and neurons during viral infection
.
Am J Pathol
 
2001
;
159
:
1219
24
45.
van
der Maesen K
Hinojoza
JR
Sobel
RA
.
Endothelial cell class II major histocompatibility complex molecule expression in stereotactic brain biopsies of patients with acute inflammatory/demyelinating conditions
.
J Neuropathol Exp Neurol
 
1999
;
58
:
346
58
46.
Omari
KM
Chui
R
Dorovini-Zis
K
.
Induction of beta-chemokine secretion by human brain microvessel endothelial cells via CD40/CD40L interactions
.
J Neuroimmunol
 
2004
;
146
:
203
8
47.
Quandt
J
Dorovini-Zis
K
.
The beta chemokines CCL4 and CCL5 enhance adhesion of specific CD4 + T cell subsets to human brain endothelial cells
.
J Neuropathol Exp Neurol
 
2004
;
63
:
350
62
48.
Galea
I
Bernardes-Silva
M
Forse
PA
van
Rooijen N
Liblau
RS
Perry
VH
.
An antigen-specific pathway for CD8 T cells across the blood-brain barrier
.
J Exp Med
 
2007
;
204
:
2023
30
49.
Cserr
HF
Knopf
PM
.
Cervical lymphatics, the blood-brain barrier and the immunoreactivity of the brain: A new view
.
Immunol Today
 
1992
;
13
:
507
12
50.
Goldmann
J
Kwidzinski
E
Brandt
C
Mahlo
J
Richter
D
Bechmann
I
.
T cells traffic from brain to cervical lymph nodes via the cribroid plate and the nasal mucosa
.
J Leukoc Biol
 
2006
;
80
:
797
801
51.
Owens
T
Renno
T
Taupin
V
Krakowski
M
.
Inflammatory cytokines in the brain: Does the CNS shape immune responses?
Immunol Today
 
1994
;
15
:
566
71
52.
Lehmann
PV
.
The fate of T cells in the brain: Veni, vidi, vici and veni, mori
.
Am J Pathol
 
1998
;
153
:
677
80
53.
Marchesi
VT
Gowans
JL
.
The migration of lymphocytes through the endothelium of venules in lymph nodes: An electron microscope study
.
Proc R Soc Lond B Biol Sci
 
1964
;
159
:
283
90
54.
Engelhardt
B
Wolburg
H
.
Mini-review: Transendothelial migration of leukocytes: Through the front door or around the side of the house?
Eur J Immunol
 
2004
;
34
:
2955
63
55.
Kissel
K
Hamm
S
Schulz
M
Vecchi
A
Garlanda
C
Engelhardt
B
.
Immunohistochemical localization of the murine transferrin receptor (TfR) on blood-tissue barriers using a novel anti-TfR monoclonal antibody
.
Histochem Cell Biol
 
1998
;
110
:
63
72
56.
Dore-Duffy
P
.
Pericytes: Pluripotent cells of the blood brain barrier
.
Curr Pharm Des
 
2008
;
14
:
1581
93
57.
Toft-Hansen
H
Buist
R
Sun
XJ
Schellenberg
A
Peeling
J
Owens
T
.
Metalloproteinases control brain inflammation induced by pertussis toxin in mice overexpressing the chemokine CCL2 in the central nervous system
.
J Immunol
 
2006
;
177
:
7242
49
58.
Farina
C
Aloisi
F
Meinl
E
.
Astrocytes are active players in cerebral innate immunity
.
Trends Immunol
 
2007
;
28
:
138
45
59.
Ransohoff
RM
Kivisakk
P
Kidd
G
.
Three or more routes for leukocyte migration into the central nervous system
.
Nat Rev Immunol
 
2003
;
3
:
569
81
60.
Janzer
RC
Raff
MC
.
Astrocytes induce blood-brain barrier properties in endothelial cells
.
Nature
 
1987
;
325
:
253
57
61.
Prat
A
Biernacki
K
Wosik
K
Antel
JP
.
Glial cell influence on the human blood-brain barrier
.
Glia
 
2001
;
36
:
145
55
62.
Wosik
K
Cayrol
R
Dodelet-Devillers
A
et al
.
Angiotensin II controls occludin function and is required for blood brain barrier maintenance: Relevance to multiple sclerosis
.
J Neurosci
 
2007
;
27
:
9032
42
63.
Sumners
C
Tang
W
Zelezna
B
Raizada
MK
.
Angiotensin II receptor subtypes are coupled with distinct signal-transduction mechanisms in neurons and astrocytes from rat brain
.
Proc Natl Acad Sci U S A
 
1991
;
88
:
7567
71
64.
Li
J
Culman
J
Hortnagl
H
et al
.
Angiotensin AT2 receptor protects against cerebral ischemia-induced neuronal injury
.
FASEB J
 
2005
;
19
:
617
19
65.
Gebke
E
Muller
AR
Jurzak
M
Gerstberger
R
.
Angiotensin II-induced calcium signalling in neurons and astrocytes of rat circumventricular organs
.
Neuroscience
 
1998
;
85
:
509
20
66.
Engelhardt
B
Wolburg-Buchholz
K
Wolburg
H
.
Involvement of the choroid plexus in central nervous system inflammation
.
Microsc Res Tech
 
2001
;
52
:
112
29
67.
Dziegielewska
KM
Ek
J
Habgood
MD
Saunders
NR
.
Development of the choroid plexus
.
Microsc Res Tech
 
2001
;
52
:
5
20
68.
Betz
LA
Goldstein
GW
Katzman
R
.
Blood-brain-cerebrospinal fluid barriers
. In:
Siegel
GJ
, ed.
Basic Neurochemistry: Molecular, Cellular, and Medical Aspects
 ,
4th
Ed.
New York
:
Raven Press
,
1989
:
591
606
69.
Wolburg
H
Wolburg-Buchholz
K
Liebner
S
Engelhardt
B
.
Claudin-1, claudin-2 and claudin-11 are present in tight junctions of choroid plexus epithelium of the mouse
.
Neurosci Lett
 
2001
;
307
:
77
80
70.
Johnson
AK
Gross
PM
.
Sensory circumventricular organs and brain homeostatic pathways
.
FASEB J
 
1993
;
7
:
678
86
71.
Leonhardt
H
.
Ependym und circumventriculäre Organe
. In:
Oksche
A
Vollrath
L
, eds.
Handbuch der mikroskopischen Anatomie des Menschen
 .
Berlin Heidelberg New York
:
Springer
,
1980
:
177
666
72.
Bouchaud
C
Bosler
O
.
The circumventricular organs of the mammalian brain with special reference to monoaminergic innervation
.
Int Rev Cytol
 
1986
;
105
:
283
327
73.
Schulz
M
Engelhardt
B
.
The circumventricular organs participate in the immunopathogenesis of experimental autoimmune encephalomyelitis
.
Cerebrospinal Fluid Res
 
2005
;
2
:
8
74.
Silwedel
C
Forster
C
.
Differential susceptibility of cerebral and cerebellar murine brain microvascular endothelial cells to loss of barrier properties in response to inflammatory stimuli
.
J Neuroimmunol
 
2006
;
179
:
37
45
75.
Allt
G
Lawrenson
JG
.
Is the pial microvessel a good model for blood-brain barrier studies?
Brain Res Brain Res Rev
 
1997
;
24
:
67
76
76.
Rascher
G
Wolburg
H
.
The tight junctions of the leptomeningeal blood-cerebrospinal fluid barrier during development
.
J Hirnforsch
 
1997
;
38
:
525
40
77.
Carrithers
MD
Visintin
I
Kang
SJ
Janeway
CA
Jr
.
Differential adhesion molecule requirements for immune surveillance and inflammatory recruitment
.
Brain
 
2000
;
123
:
1092
101
78.
Barkalow
FJ
Goodman
MJ
Gerritsen
ME
Mayadas
TN
.
Brain endothelium lack one of two pathways of P-selectin-mediated neutrophil adhesion
.
Blood
 
1996
;
88
:
4585
93
79.
Gotsch
U
Jager
U
Dominis
M
Vestweber
D
.
Expression of P-selectin on endothelial cells is upregulated by LPS and TNF-alpha in vivo
.
Cell Adhes Commun
 
1994
;
2
:
7
14